The Great Proletarian Cultural Revolution, Disruptions to Education, And Returns to Schooling in Urban China* John Giles Department of Economics Michigan State University Albert Park Department of Economics University of Michigan Juwei Zhang Institute for Population and Labor Economics Chinese Academy of Social Sciences Very Preliminary: Please Do Not Cite October 2003 *The authors acknowledge grants to support field research from Michigan State University (Intramural Research Grants Program), the Chinese Academy of Social Sciences, The Ford Foundation, and the University of Michigan (Rackham Faculty Research Grant), and support for follow-up research from the Weatherhead Center for International Affairs at Harvard University (Academy Scholars Program) and the W.E. Upjohn Institute for Employment Research.The Great Proletarian Cultural Revolution, Education Disruption, And the Returns to Schooling in Urban China 1. Introduction Over the course of the 1990s observers of China’s developing labor market have noted dramatic increases income inequality in urban areas, and many researchers have emphasized that a sharp increase in the returns to education over the course of the 1990s has contributed to these observed increases in inequality.1 Data sources used for recent research, however, often lack convincing family background variables to control for unobserved ability that may cause upward bias to estimates of returns to education. Alternatively, it is also plausible that mis-measured education could be causing downward bias on the coefficient on years of education, and that actual returns to education could be even higher than those cited in recent studies. In this paper we make use of a dataset with an abundance of family background variables, proxies for school quality and direct measures of adult literacy to control for potential upward bias caused by unobserved ability. Next, we use disruptions to educational attainment that occurred during China’s tumultuous Cultural Revolution to identify returns to education that might otherwise be biased by mis-measurement.2 1Zhang and Zhao (2002) provide a useful summary of research, and results from a standard Mincer regression using repeated cross sections of the National Statistical Bureau Urban Household Survey. 2This paper contributes to a growing literature summarized by Card (2001) in which measurement of the causal effect of education on labor earnings is facilitated using supply-side factors as exogenous determinants of schooling outcomes. Instruments used in previous studies of the returns to education include quarter of birth (Angrist and Krueger, 1991; Staiger and Stock, 1997), geographic proximity to schools (Kane and Rouse, 1993; Card, 1995), changes in school systems or school leaving age (Harmon and Walker, 1995), special education subsidies for veterans (Lemieux and Card, 1998), and a national school expansion program (Duflo, 2002). Similar in spirit to our study, Ichino and Winter-Ebmer (1998) examine educational disruption to the 1930-35 birth cohorts in Austria and Germany who were of elementary and middle school age during World War II. 2Inaccurate measurement of education may arise for several reasons. First, over the last forty years China’s education system has faced dramatic changes in curriculum content, instructional format and requirements for graduation at all levels. Years of education reported in National Statistical Bureau household surveys are typically equivalent years completed, which may differ from actual years of education completed. Moreover, the value of a year of education spent learning from farmers or industrial workers during the Cultural Revolution may be somewhat uncertain. Second, the 1990s have witnessed rapid growth in correspondence universities, television based universities (dianshi daxue) and evening after-work degree granting programs. Participants in these programs consider themselves full graduates with college education (and additional years of education), yet whether completion of these programs is comparable to additional years of schooling is less certain.3 Instrumental variables estimates presented in this paper suggest a 13.1 percent return to a year of education, which is 67 percent higher than our full-specification OLS estimate of 8.1 percent. Closer inspection of possible heterogeneity in returns to education suggests, however, that our instruments might plausibly be identifying the returns to education for individuals prevented from enrollment in college due to education supply shocks. Adding some confirmation to this possibility, we provide OLS estimates, using a piecewise linear spline in years of education, that show an 11.5 percent return to education for individuals with more than 13 years of formal schooling. 3One might suspect that the share of such participants in the labor force cannot be that large, but in the random sample of urban residents used in this paper 4.5 percent of working age adults were participating in or had participated in these types of advanced degree correspondence or evening programs. This is a significant share of total working-age adults with post-secondary education. 3One previous study (Meng and Gregory, 2002b) has also attempted to use the Cultural Revolution to study the impact of education on earnings. This study is unfortunately limited in important ways by the data used. Meng and Gregory estimate a standard earnings regression using monthly wage earnings with a limited set of controls. All of the effects of the Cultural Revolution on education are based on assumed cohort differences in shocks attributed to entire urban cohorts based on year of birth. Much is lost through this approach in that it confounds differences in education quality and content across cohorts with the shock to education, and further, assumes away important variation in education shocks across urban areas. The Cultural Revolution was, in fact, a national event, but one in which many of the political campaigns, including those affecting education, progressed in a highly chaotic and decentralized manner.4 The duration and nature of education disruptions varied across cities in urban China as well as between urban and rural areas. The data source used in this study captures this variation by making use of detailed individual educational histories and a rich set of controls. We are able to construct accurate